Antibiograms image classification based on AI techniques

Due to the overuse of antibiotics, antibiotic resistance in bacteria has emerged as a significant public health concern. Various methods can be employed to determine the susceptibility of bacteria to antibacterial compounds, and artificial intelligence (AI) has proven to be highly impactful in this field, especially with the advancements in image processing and classification. This research investigates two approaches for classifying antibiotic-resistant bacteria: traditional machine learning (ML) using the Orange Data Mining tool and deep learning with Convolution Neural Network (CNN). Using an antibiotic disk image dataset, various classifier models are compared using the Orange Data Mining tool. The traditional classifier model Logistic Regression is found to be the most accurate with a 91% accuracy rate, followed by Neural Networks (NN) with 90% accuracy. Other techniques such as Support Vector Machine (SVM), k-nearest neighbors (KNN), and decision trees (DT) are also analyzed, with varying levels of accuracy. Deep learning with CNN is found to be more efficient in classification, achieving an impressive 99.99% accuracy. This study’s high accuracy results suggest that CNN could be a promising tool for the rapid and reliable diagnosis of antibiotic-resistant bacteria in the future.